Introduction

As a non-essential heavy metal to living organisms, thallium (Tl) has attracted more and more attentions because of its high toxicity1,2,3,4. Thallium occurs at very low levels in the natural aquatic environment1,2. However, anthropogenic activities such as coal combustion, mining and processing of Tl-hosting minerals lead to the release of large amount of Tl into natural water bodies, which pose a great threat to aquatic biota and human health2. To abate the health risk associated with exposure to thallium through drinking, stringent criteria for Tl concentration in water/wastewater have been established in many countries. For instance, in the United States, the USEPA has set 2 and 140 μg/L as the maximum Tl level in drinking water and wastewater discharged, respectively; in China, the limit of Tl in drinking water has been lowered to 0.1 μg/L and more stringent standard of 2 μg/L has been adopted as the discharge standard for industrial wastewater in some provinces3,4.

In aquatic environment, thallium usually exists in two oxidation states: thallous (I) and thallic (III)3. Tl(I) is considered to be very mobile and thus difficult to remove, because it generally forms most stable compounds in natural waters1,5. Therefore, in the thallium treatment domain most of the researches concerned Tl(I) removal. A variety of techniques including adsorption6,7, oxidation/precipitation8,9,10, ion exchange11,12, solvent extraction13,14, etc., have been used to treat Tl-containing water and wastewater. In comparation with other methods, adsorption has recently gained more and more attentions, due to the advantages of high efficiency, affordable cost, simple operation and little toxic sludge generation3,6. Numerous adsorbents such as carbon materials15,16, mineral materials17,18, biomass materials19, Prussian blue and analogues20,21,22, manganese oxides23,24,25,26,27 and titanium-based materials28,29,30, have been employed to remove Tl(I) from water or wastewater.

Titanium dioxide has been extensively investigated and used to remove heavy metal contaminants such as Cs(I), Cu(II), Pb(II), Cd(II), Ni(II), As(III), As(V), Cr(III), Cr(VI), U(VI) and Th(IV) from water or wastewater31,32,33,34,35, owing to its nontoxicity, affordable cost, good chemical stability and high affinity for these ions. However, little work has been done on the removal of Tl(I) with titanium dioxide. For instance, Tl(I) adsorption on anatase TiO2 (Degussa, P25) was studied by Kajitvichyanukul et al. and the maximal adsorption capacity was found to be only 6.3 mg/g under neutral pH conditions36; Asadpour et al. investigated the Tl(I) adsorption on anatase TiO2 nanoparticles synthesized via ultrasound method and found that its maximal adsorption capacity was 25 mg/g at pH 9.037; Zhang et al. evaluated the Tl(I) adsorption on commercial rutile nano-TiO2 and determined that the maximum adsorption capacity was 51.2 mg/g at pH 7.0 ± 0.338. Evidently, these well-crystalline TiO2 nanoparticles have relatively low Tl(I) adsorption capacity and are not feasible for Tl(I) removal. Therefore, it is vital and challenging to synthesize titanium dioxide with high-efficiency adsorption of Tl(I). The amorphous TiO2 may be a feasible choice because it often possesses abundant active sites, which are responsible for Tl(I) adsorption. Recently, a poor crystalline TiO2 had been prepared by a simple precipitation method in our laboratory. The as-synthesized TiO2 demonstrated a high maximal Tl(I) adsorption capacity of 239 mg/g at pH 7.0 ± 0.129, which was remarkably superior to the well-crystalline TiO2. In addition, it could be easily synthesized in large scale. This is very interesting and the amorphous TiO2 might be a potential sorbent for effective Tl(I) removal because of its high performance, good chemical stability, cost-effectiveness, facile synthesis and environmental friendliness. However, to our best knowledge, the influence of precipitant used to prepare amorphous TiO2 on Tl(I) adsorption has never been investigated. Additionally, adsorption behavior and mechanism of Tl(I) on the amorphous TiO2 have never been systemically studied.

Hence, in this study, two different precipitants (NH3·H2O and NaOH) were used to synthesize the amorphous TiO2 via a facile precipitation method at room temperature. The synthesized TiO2 was characterized with a variety of techniques. The adsorption behaviors such as kinetics, isotherm, solution pH effect and coexisting cation influence were studied in details. Additionally, removal of thallium(I) from mining wastewater and natural river water was also evaluated. Moreover, a possible removal mechanism of thallium(I) was proposed.

Materials and methods

Materials

All chemicals such as Ti(SO4)2, NH3·H2O (30%), NaOH, NaNO3, TlNO3 and nano-TiO2 (P25) were purchased from Sinopharm Chemical Reagent Co. Ltd (Shanghai, China) and were analytical grade and used without further purification. Tl(I) stock solution was prepared by dissolving TlNO3 in deionized water. Prior to use, the working solution was freshly prepared by diluting Tl(I) stock solution to specified concentration with deionized water.

Preparation of titanium dioxide

Titanium dioxide was prepared by a simple chemical precipitation method at room temperature. Briefly, 7.2 g Ti(SO4)2 was dissolved in a 200 mL deionized water. Under vigorously stirring, 10% ammonia solution or 1 M NaOH solution was then dropwise added to the Ti(SO4)2 solution until the pH was raised to approximately 7.5. The white precipitates produced were washed for several times using deionized water, then filtrated and dried at 55 °C for 24 h. The obtained titanium dioxides were denoted as TiO2I (using NH3·H2O as precipitant) and TiO2II (using NaOH as precipitant), respectively. In addition, titanium dioxide was also prepared by forced hydrolysis of Ti(SO4)2 at 70 °C for 4 h, and the as-prepared sample was denoted as TiO2III.

Characterization

X-ray diffraction (XRD) analysis was performed on a PW3040/60 diffractometer (Philips Co., the Netherlands). The morphology of the synthesized and commercial TiO2 was observed with a Sigma 500 field scanning electron microscope (FESEM) (Carl Zeiss, Germany) and transmission electron microscope (TEM) (JEM-1230, JEOL, Japan). X-ray photoelectron spectra (XPS) were collected on an AXIS Supra spectrometer (Shimadzu Co., Japan) with a monochromatic Al Ka X-ray source (1486.6 eV). The XPS results were collected in binding energy forms and fitted using a nonlinear least-squares curve-fitting program (XPSPEAK41 Software).

Tl(I) adsorption experiments

Batch tests were performed to estimate Tl(I) removal by the synthesized and commercial TiO2. Briefly, 10 mg TiO2 was added into 100 mL polyethylene bottles, which contain 50 mL Tl(I) solution with different concentrations. The solution pH was adjusted with 0.1 M NaOH and/or HNO3. The bottles were then sealed and were shaken on an orbital oscillator at 180 rpm for 24 h. Afterwards, supernatant was collected and filtered through a 0.45 µm membrane. More detailed description of adsorption tests is shown in the Supplementary Material.

Tl(I) removal from real surface water and wastewater

To estimate the practicability of the synthesized TiO2, Tl(I) removal from real wastewater and spiked surface water was studied by batch experiments. The surface water was collected from the Pearl River near to Guangzhou University, China and the mining wastewater was sampled from a mining area, Guizhou Province, China. The river water pH value was 7.56 and spiked Tl(I) concentration was 20 μg/L. More detailed parameters of water quality were listed in Table S1. The pH value of mining wastewater was 2.73 and Tl concentration was 4.9 μg/L. More detailed parameters of water quality were summarized in Table S2. For the spiked river water, defined amount of TiO2I (10 or 20 or 40 mg) was added into a 2000-mL beaker containing 1000 mL spiked Pearl River water. Afterwards, the solution was agitated by a magnetic stirrer at a speed of 200 rpm. 5 mL water sample was taken from the beaker at predetermined times. The samples were then filtered using a filter with 0.45-μm membrane. The residual Tl concentration was measured by an inductively coupled plasma mass spectrometry (ICP-MS). For mining wastewater, the test procedure was similar to the spiked river water.

Analytical methods

Before analysis, the aqueous samples collected were acidified with HNO3 solution, and stored in glass bottles. Tl(I) concentration was determined by inductively coupled plasma mass spectrometry (ICP-OES, Avio 200, Perkin Elmer Co. USA). Trace level Tl was determined using an inductively coupled plasma mass spectrometry machine (ICP–MS, NexION 300, Perkin Elmer Co. USA).

Results and discussion

Characterization of TiO2

Figure 1 shows the X-ray diffraction patterns of synthesized and commercial titanium dioxides. For TiO2I and TiO2II, no obvious diffraction peaks can be observed, indicating that both of them are amorphous. Wang et al. had also synthesized amorphous TiO2 and observed similar phenomenon39. For TiO2III, five weak peaks appear at approximately 25.2, 37.6, 47.7, 54.7 and 62.4°, respectively, which are corresponding to the characteristic diffraction peaks of anatase (PDF#21-1272). This suggests that the anatase TiO2III is not well-crystalline. For commercial TiO2, several strong peaks appear at 25.4°, 27.4°, 35.9°, 37.8°, 48.1°, 53.8°, 55.0° and 62.8°, respectively. The peaks at 25.4°, 37.8°, 48.1°, 53.8°, 55.0° and 62.8° coincide with those of anatase and peaks at 27.4 and 35.9° are in agreement with those of rutile, implying that the commercial TiO2 contains both well-crystalline anatase and rutile phase. Figure 2 exhibits the SEM images of the synthesized and commercial titanium dioxides. As can be seen, both TiO2I and TiO2II are irregular in shape and constituted by many small particles, while the TiO2III demonstrates a regular sphere-like shape with a particle size of 1–3 μm. The commercial TiO2 particles are aggregates of smaller nanoparticles. TEM images of these four materials are demonstrated in Fig. 3. The TEM images of TiO2I and TiO2II further confirm that they are agglomerates of nanoparticles and amorphous. Relatively, the TiO2III displays a polyhedron shape with certain crystallinity. The commercial TiO2 presents well-crystalline nanoparticles with particle size of about 15–30 nm.

Figure 1
figure 1

XRD patterns of synthesized and commercial TiO2.

Full size image
Figure 2
figure 2

SEM images of TiO2I (a), TiO2II (b), TiO2III (c) and commercial TiO2 (d).

Full size image
Figure 3
figure 3

TEM images of TiO2I (a), TiO2II (b), TiO2III (c) and commercial TiO2 (d).

Full size image

Adsorption isotherms

To evaluate the Tl(I) adsorption capacities of the synthesized and commercial titanium dioxides, the adsorption isotherm experiments were conducted at neutral circumstance. The results are illustrated in Fig. 4. Clearly, the adsorption capacities of synthesized titanium dioxides are far higher than that of the commercial one. Furthermore, the TiO2 synthesized via chemical precipitation has much higher adsorption capacity than the one prepared by forced hydrolysis. The differences in the maximum adsorption capacity between them might be ascribed to their crystallinity. The amorphous TiO2 may have more surface hydroxyl groups than well-crystalline TiO2, which are responsible for the Tl(I) adsorption. Interestingly, both TiO2I and TiO2II are rather efficient for Tl(I) removal, particularly for low concentration of Tl(I). In addition, it can be seen that the precipitant used to synthesize amorphous TiO2 has great effect on Tl(I) adsorption. The TiO2 obtained using NH3·H2O as precipitant is much more effective for Tl(I) adsorption. However, extensive use of NH3·H2O might lead to ammonia pollution. The experimental data were fitted by the Langmuir model (Eq. (S1)) and Freundlich model (Eq. (S2)). The fitting curves are demonstrated in Fig. 4 and the adsorption constants obtained from the isotherms are listed in Table 2. It can be observed that the Langmuir model is more suitable for describing the adsorption behavior, due to the higher regression coefficients (Table 1). This indicates that the Tl(I) adsorption on the TiO2 follows a monolayer adsorption process, since the Langmuir model assumes that adsorption is limited to one monolayer. The maximal adsorption capacities of TiO2I, TiO2II, TiO2III and commercial TiO2 are 302.6, 230.3, 106.3 and 34.7 mg/g at pH 7.0, respectively. A comparison between the synthesized TiO2 and adsorbents reported in literature for Tl(I) adsorption has been done (Table 2). Evidently, both TiO2I and TiO2II are more competitive than the majority of reported adsorbents, implying that amorphous TiO2 is a promising alternative for Tl(I) removal from water. Therefore, investigation was focused on the TiO2I and TiO2II in the following sections.

Figure 4
figure 4

Adsorption isotherms of Tl(I) on the synthesized and commercial TiO2. Experiment conditions: Adsorbent dosage = 0.2 g/L; pH = 7.0 ± 0.1 and T = 25 ± 1 °C.

Full size image
Table 1 Langmuir and Freundlich isotherms parameters for Tl(I) adsorption on the synthesized and commercial titanium dioxides at pH 7.0 ± 0.1.
Full size table
Table 2 Comparison of Tl(I) maximal sorption capacities for different adsorbents.
Full size table

Tl(I) adsorption kinetics

Figure 5 presents the adsorption kinetics data of Tl(I) on the TiO2I and TiO2II. A fast adsorption of Tl(I) was observed within the first 0.5 h. During this period, about 87.3 and 81.1% of equilibrium Tl(I) adsorption capacity was achieved for TiO2I and TiO2II, respectively. Afterwards, the Tl(I) adsorption rate became slower and the equilibrium was established within about 4 h. Both the pseudo-first-order model (Eq. (S3)) and pseudo-second-order model (Eq. (S4)) were initially applied to stimulate the kinetic data. The fitting curves are depicted in Fig. 5 and constants obtained from these two models are provided in Table 3. In terms of R2, it has been found that the pseudo-second order model fits better the kinetic data than the pseudo-first order model, indicating that removal process of Tl(I) by the synthesized TiO2 involves chemisorption.

Figure 5
figure 5

Kinetics of Tl(I) adsorption on TiO2I (a) and TiO2II (b). Experiment conditions: Initial Tl(I) concentration = 34.5 mg/L; adsorbent dosage = 0.2 g/L; pH = 7.0 ± 0.1 and T = 25 ± 1 °C.

Full size image
Table 3 Kinetic parameters for Tl(I) adsorption on the TiO2I and TiO2II fitted with the pseudo first order and pseudo second order models.
Full size table

Adsorption process is complicated and multi-step, involving bulk diffusion (adsorbate transport from the bulk solution to the outer surface of the liquid film), film diffusion (from the outer surface of the liquid film to the surface of the solid adsorbent), intraparticle diffusion (from the surface of the adsorbent to the interior pores), and adsorption on the surface actives of solid adsorbent46,47. The pseudo-second order model is therefore limited in accuracy because it considers adsorption as a single, one-step binding process48. Thus, intraparticle diffusion model (Eq. (S5)) was further used to describe the experimental data. The fitting results are shown in Fig. S1. For both TiO2I and TiO2II, the plot of qt vs t1/2 can be divided into two linear segments, indicating that Tl(I) adsorption contains multiple steps. The first linear section is corresponding to the fast adsorption stage, which is mainly controlled by film diffusion. The second linear section is a slow stage, which is governed by the diffusion of Tl(I) from the surface of adsorbent into the micropores.

Influences of pH and ionic strength on Tl(I) adsorption

The solution pH affects not only the species of metal ions but also the surface functional group on the adsorbents for metal ions capturing49. Figure 6 demonstrates the influences of solution pH and ionic strength on Tl(I) adsorption. Evidently, Tl(I) adsorption on the TiO2I and TiO2II is strongly affected by the solution pH value, increasing gradually with its increase (2.0–9.0). And the optimal adsorption occurs under high alkaline circumstances. A similar trend was observed for the Tl(I) adsorption on titanium peroxide29 and titanium iron magnetic adsorbent43. In the tested pH range from 2.0 to 9.0, positively-charged Tl+ is the dominant species for Tl(I). Under acidic conditions, the surface of TiO2 was favorably protonated and consequently positively charged, which resulted in strong electrostatic repulsion between Tl+ and the positively-charged surface and depression of Tl+ sorption. With the increase in solution pH value, the TiO2 surface became less positively-charged and turned to be negatively-charged, which was beneficial for the sorption of Tl+. Thus, the uptake of Tl+ increased.

Figure 6
figure 6

Effects of solution pH and ionic strength on Tl(I) adsorption by TiO2I (a) and TiO2II (b). Experimental conditions: Initial Tl(I) concentration: 30 mg/L; adsorbent dosage: 0.2 g/L; T: 25 ± 1 °C.

Full size image

As can be seen in Fig. 6, the change in ionic strength (from 0.001 to 0.1 mM) did not greatly affect the adsorption of Tl(I) on both TiO2I and TiO2II. Adsorption of ions by formation of outer-sphere complexes is very sensitive to the ionic strength change and always decreases with an increase in ionic strength, since the background electrolyte ions can also form this kind of complex via electrostatic force. On the contrary, adsorption by formation of inner-sphere complexes is insensitive to the variation of ionic strength50. Thus, it could be reasonably concluded that the Tl(I) was specifically adsorbed on the surface of TiO2 by formation of inner-sphere complexes.

Influence of coexisting cations

Cations such as Ca2+, Mg2+ and K+ often exist in the surface water and groundwater51,52. Moreover, heavy metal ions such as Zn2+, Ni2+ and Cd2+ co-occur frequently with Tl+ in the mining and industrial wastewaters. These present cations might compete for the adsorptive sites on the surface of TiO2I or TiO2II with Tl+. Therefore, the influence of these cations on Tl+ adsorption was evaluated by batch tests at pH 4.5 ± 0.1.

Figure 7 shows the experimental results. Interestingly, the coexisting K+, Ca2+, Mg2+, Zn2+ and Ni2+ do not greatly affect the Tl+ adsorption and no great decrease is observed even though the concentration of present cations is as high as 10 mM. Awual et al. also found that the present K+ did not greatly prevent the adsorption of Cs+ (similar to Tl+ in chemical properties) by crown ether based conjugate material53. It is noteworthy that the coexisting Cd2+ inhibits Tl+ adsorption. For the TiO2I, this negative effect is slight and the Tl+ adsorption capacity still remains over 90% when the concentration of Cd2+ reaches up to 10 mM, being 100 times higher than that of initial Tl+. However, for the TiO2II, the negative influence is relatively remarkable and T(I) adsorption decreases by about 29% as the concentration of Cd2+ increases from 0 to 10 mM. Relatively, both TiO2I and TiO2II has high selectivity towards Tl(I) in the presence of multiple cations.

Figure 7
figure 7

Influence of coexisting cations on Tl(I) adsorption by the TiO2I (a) and TiO2II (b). Experiment conditions: Tl(I) concentration = 18.5 mg/L; adsorbent dosage = 0.2 g/L; pH 4.5 ± 0.1 and T = 25 ± 1 °C.

Full size image

From above, we can see that the adsorption behaviors of Tl(I) on TiO2I and TiO2II are very similar. Therefore, we only studied the applicability, regeneration and mechanism of Tl(I) removal by the TiO2I in the following sections.

Tl(I) removal from real surface water and wastewater

To evaluate the applicability of the TiO2I, kinetics of Tl(I) removal from the surface water (spiked Pearl River water) and mining wastewater were respectively investigated by batch tests. The results are shown in Fig. 8a,b, respectively. For the Pearl River water, when the dosage of TiO2I was 20 mg/L, the concentration of residual Tl(I) in the effluent was lowered to less 2 μg/L within 240 min (Fig. 8a). Tl(I) removal became more rapid as the dosage of TiO2I increased. When the dosage was 40 mg/L, the residual Tl(I) reduced to less 2 μg/L within 120 min and below 1 μg/L at 360 min. For mining wastewater, when the dosage was 25 mg/L, the Tl concentration in the effluent was below 2 μg/L after treatment for 210 min. As the dosage increased to 50 mg/L, the residual Tl decreased rapidly from 4.7 to less 2 μg/L within 30 min (Fig. 8b). These results suggest that the TiO2I is highly efficient for Tl(I) removal from the river water and real mining wastewater and has good applicability.

Figure 8
figure 8

Kinetics of Tl(I) removal by TiO2I from the (a) spiked Pearl River water and (b) mining wastewater.

Full size image

Regeneration and reusability of TiO2 I

The regeneration and reusability of adsorbent is an important factor affecting its use in real water treatment54. In order to assess the reusability of TiO2I, the Tl(I) desorption from spent adsorbent was investigated using 0.1 M HCl solution as desorbing agent and then the regenerated adsorbent was used in another adsorption–desorption cycle. Figure 9 illustrates the results of five consecutive adsorption/regeneration cycles. The cycle 0 is corresponding to the Tl(I) adsorption by the fresh TiO2I. As can be seen, the adsorption percentage of Tl(I) decreases with an increase in the number of cycles. After the first regeneration, the adsorption percentage of Tl(I) by the regenerated adsorbent reduces from 98.7 to 79.3%. This value is further lowered to 60.1% after the third regeneration and 45.3% after the fifth regeneration. Apparently, the reusability of TiO2I is moderate, which may be ascribed to the relatively strong affinity between Tl(I) and the TiO2I. These results suggest that the TiO2I could be regenerated but the times of reuse are limited.

Figure 9
figure 9

Variation of Tl(I) adsorption by the TiO2I as a function of regeneration cycle.

Full size image

XPS analysis before and after Tl(I) adsorption

In order to reveal the mechanism of Tl(I) adsorption by the TiO2I, XPS spectra of the TiO2I before and after Tl(I) uptake were determined and analyzed. Figure 10a presents the survey spectra of the original and Tl(I)-sorbed TiO2I. Characteristic Ti peaks including Ti 2p, Ti 2s, Ti 3p, Ti 3s and Ti KLL along with O peaks are observed in the spectra of the original TiO2I. After reaction with Tl(I), two characteristic Tl peaks of Tl 4f and Tl 4d appear, suggesting that Tl(I) was adsorbed on the surface of TiO2I. High resolution XPS spectra of Ti 2p, Tl 4f and O 1s of the pristine and Tl(I)-loaded TiO2I are illustrated in Fig. 10b–d, respectively. The doublet peaks of Ti 2p3/2 and Ti 2p1/2 are located at 458.7 eV and 464.4 eV, respectively, indicating that the oxidation state of Ti in the TiO2I is + 429,55,56. These two peaks exhibit a slight shift (0.2 eV) to lower binding energy after Tl(I) sorption, which might be ascribed to the presence of strong interaction between TiO2I and Tl(I). The two peaks of Tl 4f7/2 and Tl 4f5/2 are located at 119.1 eV and 123.5 eV, respectively, indicating that the oxidation state of Tl sorbed is + 123. Obviously, no Tl(I) oxidation occurs during its adsorption by TiO2I. The O 1s spectra can be divided into three peaks situated at 530.2, 531.7 and 533.1, corresponding to lattice oxygen (O2−), surface hydroxyl (–OH), and sorbed water (H2O), respectively. For the virgin TiO2I, the contents of O2−, –OH and H2O are 63.3, 28.9 and 7.8%, respectively. After Tl(I) sorption, the content of H2O showed no significant change, while the content of –OH species decreased obviously from 28.9 to 17.2% and correspondingly, the content of O2− increased from 63.3 to 75.1%. Obviously, the H+ in –OH group was replaced by the Tl(I) species during its removal.

Figure 10
figure 10

XPS spectra of TiO2I before and after Tl(I) adsorption. (a) Survey spectra, (b) high-resolution Ti 2P spectra, (c) high-resolution Tl 4f spectra, (d) high-resolution O 1s spectra.

Full size image

From the above-mentioned analysis, a possible mechanism of Tl(I) removal by the TiO2I was established and the schematic diagram was illustrated in Fig. 11. Firstly, the Tl+ was transported to the surface of TiO2I from bulk solution. Afterwards, the Tl+ replaced the H+ in –OH group on the surface of TiO2I and an inner-sphere surface complex (Ti–O–Tl) was formed. Meanwhile, the H+ was released and entered into the bulk solution.

Figure 11
figure 11

The proposed mechanism of Tl(I) adsorption on TiO2.

Full size image

Conclusions

Hydrous titanium dioxide was facilely synthesized by precipitation method and forced hydrolysis method, respectively. The TiO2 prepared at room temperature is amorphous and effective for Tl(I) adsorption, exhibiting high maximal adsorption capacities of 230.3–302.6 mg/g under neutral pH conditions. These values outperform the majority of reported adsorbents. The Tl(I) adsorption is strongly pH-dependent, increasing with an increase in solution pH value. The TiO2 has high selectivity for T(I) adsorption and it can be used repeatedly, though the times of reuse are limited. The mechanism of Tl(I) removal is that the H+ in –OH on the surface of TiO2 was replaced by Tl+ and inner-sphere surface complex was formed. The synthesized TiO2 has the potential to be used as an alternative adsorbent to remove Tl(I) from water, owing to its high efficiency, high stability, affordable cost, facile synthesis and environmental friendliness.